Streaming OpenAI response

I am currently deploying my streaming code to production and am concerned about potential issues with out-of-order data. Can anyone advise on whether this is a significant concern and how to mitigate it?